Two Think Minimum Podcast Transcript
Episode 004: “Incentive Auctions and Spectrum Thoughts”
Recorded on February 13, 2018
Chris: 00:00 Hello, and welcome to another episode of TPI’s podcast, Two Think Minimum. I’m Chris McGurn, TPI’s Director of Communications. On this podcast, we have focused on topics ranging from the definition of broadband to infrastructure broadly and what it means for telecommunications, the Universal Service Fund and rural broadband access. Today, we will continue to tackle the meta issues in tech policy and tech politics with a conversation on the Incentive Auctions.
The last incentive auction held on March 30, 2017 yielded $19.8 billion in revenue, including $10.05 billion for winning broadcast bidders and more than $7 billion for the United States Treasury.[1] Now, for those of us not familiar with the technical side of this policy, when it comes to incentive auctions and spectrum in general, these terms may be new and it might be staggering that they’ve raised so much money, but that’s why we’re joined by Scott Wallsten, who is President and Senior Fellow of TPI, and Sarah Oh, fellow at TPI, to delve into topics all things incentive auctions today. I’ll let them take it away.
Sarah: 01:14 Scott, I know that you did some work on researching the incentive auctions and discussing alternatives to running the incentive auction, but also interpreting results from it. I recently read a blog post that TPI published comparing auctions and auction revenues from the last few spectrum auctions. The AWS-3 spectrum auction yielded $41 billion in revenues, which was an outlier,[2] with a high MHz-POP number over $2 /MHz-POP. In comparison, the incentive auction from spring 2017 yielded revenues of $19 billion, which was less than half of the AWS-3 auction, but also with lower MHz-POP averages at around $1.30 /MHz-POP for top 40 markets. Do these numbers mean that the incentive auction of the 600 MHz band wasn’t as successful as the AWS-3 auctions? Did it meet expectations?
Scott: 02:29 Let’s back up a little bit. The short answer to your question is that the auction was a success. It did not meet many expectations, but those expectations were unrealistic because they were based on the previous the AWS-3 auction, which was, which yielded numbers like we’d never seen before. If you look at expectations prior to that auction, then the incentive auction pretty much met them. The incentive auction did what it was supposed to do and the FCC deserves a lot of credit for it. Now that doesn’t mean there aren’t a lot of things we can fix it and there’s still a debate about whether it’s the right thing to do. But let’s back up even further. Let’s sort of slowly work backwards.
When you gave those prices, you said that they were in terms of MHz-POP, or price per MHz-POP. So you need to explain what that is. That’s the price – the dollars that was paid for a given license – times the bandwidth of the license, divided by the population covered by the geographic footprint. That’s the standard way of normalizing prices of spectrum so that we can compare them.[3]
We didn’t always use to use auctions at all. Starting in the 1930s, and probably earlier before the FCC, spectrum was allocated by what we call “beauty contests.” There would be spectrum available and the FCC would review arguments for what it should be used for and who should get it. They would decide, let’s say the 600 MHz band of spectrum should be used for television broadcast. Then anybody could come in and make a case for why they were the ones that deserved it.
They had spectrum for all kinds of things, for various specific use cases, for TV, for radio, for taxi, taxi drivers, and taxi cab companies, for all kinds of things. The problem with that was that, first of all, we didn’t know whether that was really the best use because everybody’s going to come in and argue that what they do is the best use and people still do that actually. If somebody is allocated spectrum today, you will always hear them forever say that they are the ones that need that spectrum. We don’t know whether it was initially allocated properly and then because it was mandated for that use, it was hard for it to switch to something else when that was no longer the most valuable use of the spectrum.
They tried to get rid of the problem of the “beauty contest” of everyone trying to convince them it was better by having it be a lottery system and relying on the secondary market to move that spectrum to its highest value use. The problem with that was that the transaction costs, um, led to people earning windfalls. And nobody likes windfalls. Economists don’t care about windfalls because those aren’t real economic effects. They’re just transfers of money and so if the secondary market moved the spectrum to its highest value use and somebody made a fortune on it, well, it was lucky for that person and because it belonged to the government, let’s say, it was publicly owned beforehand, arguably that money should have gone to the government, but it didn’t affect the economic output of that.
Eventually, thanks to people like Evan Kwerel, an economist at the FCC who had pushed for auctions and done research on it for a long time, and Reed Hunt who was chairman of the FCC at the time and knew to listen to his staff, we ended up bringing in auctions. The idea was that we could figure out where the spectrum would make the most sense, who had the highest value use, who was willing to pay the most for it. We auctioned it off just like we would auction off anything else, or issue requests for proposals and give it to whoever was willing to pay the most. That’s called a forward auction, a standard auction, what people think of where people bid higher and higher prices until no one was willing to pay anymore.
Scott: 06:35 The incentive auction was a way to simultaneously free up spectrum that was stuck in this antiquated use case and then turn around and auction it off. In this case it was a way to get spectrum from the broadcasters by seeing how much they would have to be paid to give up their license rather than just demanding that they give it up, and allowing a broadcasters to opt in if they wanted to whoever on TV stations as they could get a payout. Then they would repack the existing stations to create a nice clean block of spectrum and turn around and auction it off. That was the incentive auction. I think I forgot your question after all that.
Sarah: 07:21 Thanks for the history on spectrum.
Scott: That sounded very sarcastic.
Chris: Still a lot of good information though.
Sarah: When you were recounting the history of spectrum, I just keep wondering, “Why is it so political?” It sounds so simple when you say, “the FCC held an incentive auction,” but getting to auctions took a long time, a lot of politics, a lot of congressional involvement, a lot of persuasion, a lot of equity arguments. It’s not a simple process and you would think that something as simple as spectrum, I mean as “simple” like a resource like spectrum could be moved around with market mechanisms.[4] So why is Congress so involved? The incentive auction was only authorized by the Middle Class Tax Act in 2012,[5] ten years after a report from the FCC recommending this incentive auction by Evan Kwerel [and John Williams].[6]
Scott: 08:32 You’re asking really interesting questions. Why spectrum is so political? One reason is that it’s so valuable, but lots of things are valuable and they’re not, they’re not political. Partly, it goes back to a few things. One is that it was allocated to certain uses, so lots of groups feel like they have the right to particular bands of spectrum and they don’t want to give it up. That also includes not just, like we were talking about the TV broadcasters, but the government – the Defense Department – is the largest single holder of spectrum and arguably doesn’t use it very efficiently and we’d like to move more of it for private and civil use, private civil commercial use. They don’t want to give it up. Nobody wants to give up something that they feel belongs to them. We have to figure out ways of doing that that everyone agrees with, or not everyone, but enough people agree with. That creates constant arguments. There are groups that would much rather fight about it in a regulatory setting than have to pay for it.
Sarah: 09:43 To be a squeaky wheel, can I interject here? Sometimes I wonder, the unlicensed spectrum crowd, they are just repeating the same mistakes that led to the broadcasters having spectrum and making it political. If unlicensed bands are just regulated by rules that go through the FCC,[7] what happens when we need to clear devices off of unlicensed? It has to go through this whole political process again. Even though in the short term it sounds like, having unlicensed rules is the fastest way to have new uses enter the spectrum quickly, in a bigger picture, I feel like licenses that are able to move around in the market are actually better for innovation.
Scott: 10:32 First, since we don’t have somebody here who is specifically pro-unlicensed, we don’t want to criticize them unfairly or put words in anyone’s mouth even though it’s fun. Unlicensed is important, and licensed is important. But you’re raising the point that, when a band of spectrum is allocated for a particular use, it becomes very, very difficult to change it. Now, unlicensed is a little bit different from just specifying a use, because it can be any use as long as it meets certain rules regarding interference and accepting interference and so on. Unlicensed has been important for innovation, we use Wi-Fi all the time.
But we did see it when there was a new technology that wanted to use unlicensed – LTE-U – caused a huge fight. The Wi-Fi proponents didn’t like it. Then there were lots of technical arguments over, did it cause interference and, and how much.[8] On the one hand, the rule is very clear – if it causes interference, it doesn’t matter which, it’s not supposed to matter, anybody can use it. But there was a huge installed base of people who used Wi-Fi and so it became its own constituency. Broadly speaking that is a problem.
One of the problems that we’re faced with now is that cars, automobiles, were given spectrum in the 5 GHz band and now they are fighting tooth and nail to make sure they keep that spectrum. It doesn’t matter whether they’re actually going to use it, they say they need it for safety reasons. So, here’s another industry that was given access to have, what it considers its spectrum, and doesn’t want to give it up. That’s why we try not to give spectrum to groups because it becomes a constituency that doesn’t want to give it up because there is no easy way to compensate.
Sarah: 12:25 Why isn’t spectrum like any other property? What makes spectrum so political? Scott, you talked a little bit about that. The initial allocations, there’s a lot of history involved, like back in the Hoover administration where the federal government was organizing “chaos on the airwaves.”[9] There was a big increase in the number of radio stations that were broadcasting. Then, they were asking the federal government to put some order on entry because there was just too much interference. That’s an example of where the federal government served as an arbiter or moderator in the market.
Scott: 13:12 On the one hand, it was important for dealing with the interference. On the other hand, there were a group of incumbents looking to be protected from entry.
Sarah: 13:20 It’s like an area where it’s uniquely – Tom Hazlett says this in his book too – where it’s uniquely convenient to come up with a technology reason for obstruction, or technology reason for a carve out from the regulator or from Congress.
Scott: 13:37 We’ll have Tom as a guest on the podcast. One of the most insightful things from his book, The Political Spectrum,[10] is that wherever he went, the objection to making spectrum available for some other use, or more freely available, the objections were always technical. You have sort of a rule of thumb when somebody makes a technical argument against, allowing use – you can bet that they’re just protecting an incumbent, and that’s the reason. There’s always a political economic reason and the technical stuff isn’t the real issue. I’m sure that’s exaggerating because there are technical issues.
Sarah: 14:26 Ronald Coase also – he has such good economic instincts – gravitated towards studying spectrum and the FCC, noticing that there’s just so much economics and incentives in the whole movement of licenses and rights on spectrum, far more than technical constraints.[11]
Scott: 14:51 He realized that the beauty contest way of allocating spectrum was silly and he wanted a system of auctions essentially. And he was laughed at originally.
Chris: You talk about some of the other ways spectrum was allocated, the beauty contest and the lottery, and then moving into an auction system where it goes to the highest bidder, doesn’t that block out small innovative companies if they wanted to get spectrum?
Scott: 15:23 Building out a wireless network is hugely capital intensive. The idea that a small company is going to build its own infrastructure is not especially realistic. The FCC has tried to give small company credits in the auctions, but mostly, that just creates shell companies and all kinds of ways of gaming the system. There’s no evidence that it’s done any good. Plus, we want innovative uses on those wireless networks. I mean we also want the wireless networks themselves to be innovative. But set asides in the infrastructure for small businesses, is generally not a good way to go.
Chris: 16:10 When we’re talking about the auctions themselves, we’re really talking about a handful of companies that everyone knows about because they pay their monthly wireless, Wi-Fi, cellular bill to them. Are those the players when we’re talking about in the spectrum auctions?
Scott: 16:23 Usually, but not always. There are, in any given auction, companies that you probably haven’t heard of. Then there are companies that you’ve heard of, that are in the auctions, surprisingly, in the auctions. In the AWS-3 auction, DISH was in it and obviously it sounded like a company that spectrum is important to them, but their bidding behavior caused the prices to go much higher and so, made the usual guys, AT&T, Verizon, pay a lot more for that spectrum than they would have otherwise. Sometimes Comcast has been in the auctions. Google was in an auction. They’re a good number of companies that participate. Although the companies with the largest collections of spectrum are, of course, the biggest wireless carriers.
Chris: 17:16 When you talk about companies like DISH and Comcast, and Google, being in them, are they basically pulling a Bart Simpson where they’re going in and trying to raise the price just to cause a little bit of pain for their competitors or what was their justification?
Scott: 17:29 Well, that’s a pretty risky strategy because you might end up winning and then you owe billions of dollars. That said, Charlie Ergen is hard to figure out. I say that, not as criticism, because he’s made billions of dollars by being that way, so he knows what he’s doing. Even if we don’t.
Sarah: 17:52 DISH bid on 600 megahertz spectrum, I think they were one of the top bidders.
Scott: 17:56 But sometimes, with Comcast, at one time some years ago, there was a consortium of cable companies that bid for spectrum thinking that that would be part of their wireless strategy and that they might launch their own network. It didn’t pan out and they sold that spectrum to Verizon. I believe most of it, I think not all of it, went to Verizon. They participated in the auction because they thought they were going to need that spectrum as an input into a service. It turned out they didn’t, it went a different way. They started going down the Wi-Fi route.
Sarah: 18:35 I was browsing through the winners of the incentive auction.[12] There are a lot of private equity groups that are buying up smaller chunks of the spectrum. I’m not sure what they do with it. They resell it, I don’t know. But there are a lot more bidders than I would expect. It’s way more than the top four.
Scott: 18:55 We know that we should keep moving more spectrum into the market because the price for it is, positive. We know it has value, but if it were, if there were increasing demand, you would expect to see spectrum prices continuously increasing. That’s not the case. Sometimes it increases, sometimes it decreases. It’s different by different bands. The reason for that, is that wireless demand for wireless services is growing by leaps and bounds. We know that, like Sarah said, spectrum is one input and the other inputs are basically technology, and technology improves. We use spectrum more and more efficiently all the time. As wireless demand increases, if you’re running a network, you have choices on how to expand, you could get more spectrum or you can improve the technology you’re using, you can split cells, you could do all kinds of things. At any given moment, you’re going to make a trade-off between investing in technology or infrastructure, and getting more spectrum. The value of the spectrum is going to depend not just on demand for wireless services, but on how quickly, and the price and cost of new technology and infrastructure.
Sarah: 20:08 One thing that I find interesting is the value of unencumbered spectrum or even, paired spectrum, is far greater than –
Scott: Tell people what paired and unpaired means.
Sarah: Paired spectrum is when you can get two channels that are paired together, an uplink and downlink. Whereas unpaired spectrum is a band that is just by itself. And, I believe, paired spectrum is more valuable on the market because there’s more demand.
Scott: 20:38 Paired spectrum is more valuable because it uses a technology that’s much more common –
Sarah: 20:42 Right, and because there are subscribers on the other end of the network, so this is an area where the market drives the value of spectrum. It doesn’t have real a value on its own, like anything really. It’s the price that is set when the market demands it. Another thing about unencumbered spectrum versus encumbered spectrum, there’s a debate on whether spectrum should be shared or not with multiple users, or on a primary or secondary basis. But I believe studies have shown that unencumbered spectrum, at least for mobile carriers, is valued more highly, so where you are the primary user, you don’t have to worry about interference or sharing. Is that true?
Scott: 21:32 There are different kinds of encumbrances though. Some can make the spectrum unusable until you clear it. Then of course it’s less valuable because it’s going to cost you to clear the spectrum and it’ll take time before you can start already earning money on it.[13] That was one of the justifications for doing the incentive auction that by doing the reverse auction first, then repacking the remaining stations, the FCC could then auction off clear and open blocks of spectrum that were unencumbered. The harder it is to use your spectrum, the less valuable it’s going to be, all else equal.
Chris: 22:13 When we’re talking about the beach front property, spectrum, do we have any idea, any estimates on how much that would cost on the auction market?
Scott: 22:24 We can go and look at the data. Every auction has been different. Generally speaking, until now, lower frequency spectrum, meaning 600, 700 MHz spectrum, is more valuable than spectrum higher up like, 1.9 GHz, 2.5 GHz spectrum. Now we hear a lot about 5G these days and that will use higher frequency spectrum. We don’t know yet how we will value that. We haven’t seen an auction for it yet. So, you know, it could turn out to be all hype and it’ll turn out that that higher frequency spectrum is much less valuable in a normalized sense then the lower spectrum and it’ll be the same trend as always, or maybe the demand for densification, which is what 5G will be about, adding more capacity to where there’s already coverage, maybe they’ll turn out to be very high demand for that and we’ll see, the demand for the spectrum outstrip the infrastructure and we will see higher prices. But we don’t have evidence on that yet.
Sarah: 23:43 Going back to alternatives to incentive auctions. Incentive auctions are, I would say, a great innovation for the FCC, to be more market-oriented. The same thing with reverse auctions in the Universal Service Fund. But there are alternative ways of reallocating resources while also coming to compromise. Should we talk a little bit about alternatives to incentive auctions? Basically, the basic problem is reallocating existing incumbent users, without giving them a windfall, which some people are concerned about, but also moving spectrum to higher valued uses and liberalizing the rules. Like we said before, turning a license that’s only for broadcast TV use into a general purpose license, isn’t as simple as just the FCC saying so. I wish it would be.
Scott: 24:49 It actually can be as simple as that. The question is, is whether then, the transaction costs would be too high then for anybody to reassemble that into useful spectrum. That’s an ongoing debate.
Sarah: 25:01 Ronald Coase would probably say that’s fine, right?
Scott: 25:05 He would say that if there are no transaction costs then it’s fine. But he would also say that it’s the transactions costs that have shaped the market the way it is. The amount of spectrum that you need as a distributor or as an infrastructure company, does depend on what you want to do with it. You need a lot more spectrum to deliver video than you do for some Internet of Things activities.[14] Often an Internet of Things thing just needs to let you know that something’s there, right? That’s just a ping, sending a bit down a pipe or through the airwaves. That doesn’t take much spectrum at all. If you want to watch your house on HD video, 24 hours a day, that’s going to take a lot of spectrum. Sorry, I shouldn’t put it that way. It’s going to take a lot of bandwidth, and however that bandwidth is provided, whether it’s by wireless services, which will then take a lot of infrastructure and bandwidth, or wired connections that requires sufficient bandwidth. It depends a lot on the use.
Sarah: 26:13 Now that we’re talking about spectrum more generally, maybe we can wade into the unlicensed-licensed debate. Part of this conversation is that a lot of data is being offloaded through Wi-Fi and Wi-Fi is pretty much a connection to a wired cable subscription. Wouldn’t you say Wi-Fi is an extension of wired?
Scott: 26:45 A lot of people say that all spectrum is an extension of wired, because most backhaul to finally take the connection back to a server somewhere, there’s a wire somewhere. It’s always bringing it back to a cable somewhere. Wi-Fi is a shorter distance to the backhaul, definitely. Maybe even backhaul isn’t the right word for Wi-Fi because it’s taking it to your home wire connection or your work wired connection and so on. But people, just in general use are often confused about what wireless means because wireless just means wireless. All people care about is whether they can carry their device around. They don’t care whether it’s connecting to Wi-Fi or cellular or a 600 MHz or 1.9 GHz, except for maybe us who like just knowing very strange references.
Sarah: 27:45 The term, “mobility,” can often refer to the wireless network, but Wi-Fi is part of your home connection. There is kind of a debate on, “Do we need more unlicensed spectrum for W-Fi uses, or do we need more licensed spectrum for mobility?”, and because national networks need to organize their communications on unencumbered licensed bands. What do you think about unlicensed and licensed in that light?
Scott: 28:25 They’re both obviously important, but part of the problem with the debate is, when people say, we need more unlicensed, it’s very hard to compare it to licensed, because whenever anything is free, you always need more of it. People love to consume things that have a price of zero. You’ll see usage on it increase quickly. On the one hand, that’s been good for Wi-Fi too, right? Because we’ve seen lots of new devices come on. We should probably, rather than just saying Wi-Fi, say, unlicensed in general, because there are other technologies, there’s LTE-U, which has been a big debate with Wi-Fi. There’s Z-Wave technology for in your house. The lock on our front door communicates with that device that can then tell me whether the door is locked or not, on my phone. I’m not sure why I need that, but I like it. It becomes just a regulatory debate, without a lot of information to help inform whether an additional megahertz of spectrum should be licensed or unlicensed because it’s hard to compare what the marginal value of the marginal megahertz is in one versus the other.
Sarah: 29:44 In economics papers, I know that this is an active area of research. There are only a handful of papers, I would say that purport to measure, first of all, economic growth that comes from one or the other, unlicensed or licensed spectrum, or even value the output from unlicensed or licensed.[15] Don’t you think these are important quantitative facts to know, before going through the policy debate?
Scott: 30:16 Absolutely. But a lot of these papers – I don’t believe papers that put the economy on the left-hand side – you just cannot measure the effects of one thing to the entire economy. In general, when you see that something added “x billion dollars” to the economy, you know it’s probably a position paper. It’s hard to know exactly what the outcome is that you want to measure and, and how to measure it. So, it’s a tough debate.
Sarah: 30:56 Right now, at least in the spectrum debate,[16] it seems like well, maybe it’s always been, evenly matched between the licensed and unlicensed crowds. I personally think,[17] licensed – liberal licenses[18] – that can move around from one innovative company to another, makes more sense than unlicensed bands that require the FCC to change the rules. But other people have different opinions.
Scott: 31:29 We’ve also seen that any incumbent – incumbents act like incumbents – and the Wi-Fi crowd was really upset about LTE-U, which was another unlicensed use. The rules are very clear on unlicensed has to accept interference from any other service in that band, but they didn’t want it. They didn’t really like the unlicensed aspect of it when it came to competition. One of the problems with unlicensed is that, you’ll always see fights at the regulator, rather than are in the market. That’s one reason why, that’s one point in favor of, licensed. It also depends on the uses. Utilities for example, don’t want to use unlicensed spectrum because they’re worried about security and they want to have more control over the network and the airwaves that they’re using.
Chris: 32:28 In the few minutes we have left, what are the one or two key takeaways that you want people who might not be as familiar with spectrum and incentive auctions to know?
Scott: 32:37 One key takeaway is that the FCC ran a really complicated auction, a reverse auction and forward auction, and that’s a big accomplishment. People who think that the government is always incompetent, should take a look at that because it involved machine learning, sophisticated auction design that hadn’t been done before, and that’s going to have implications for our future, things like universal service and other possibilities of getting back spectrum. The other thing though is that even though it was successful and admirable, we still don’t know if that was the best way to go about it because there are other ways we might have tried to repurpose spectrum. We still don’t have the empirical answer to what is the best way to do that.
Sarah: 33:30 In my mind, is that a feature or a bug that the incentive auction was so complicated? Part of it is a little bit of smoke and mirrors. Like, we’re going to do this fair auction with market prices and everyone will be happy, but there’s a lot of political calculus involved in making it complicated. If it’s too straightforward – if, I mean, the FCC could just say, these broadcast licenses will now be liberal licenses, and broadcast stations, you can now use the spectrum or sell it – In one order – and you wouldn’t need all of that [incentive auction machinery] –
Scott: 34:07 Right. The FCC’s counter to that is that, the transaction costs are too high for any given company to buy up all these disparate TV networks and try to put together a coherent footprint in the similar band. But that is the question. Did we need this complicated auction? We really don’t know the answer to that.
Sarah: We don’t know.
Chris: 34:33 We can leave it right there, we don’t know the answer, but we have given a lot of good background and advice on how to move forward. We hope you enjoyed this episode of Two Think Minimum and hope you join us again next week when we talk about Artificial Intelligence and IoT in preparation for our big conference on February 22. See you next time.
Sarah: 34:54 Bye.
[1] https://auctiondata.fcc.gov/public/projects/1000
[2] https://techpolicyinstitute.org/2017/01/17/the-fccs-incentive-auction-is-not-a-disappointment/
[3] https://techpolicyinstitute.org/wp-content/uploads/2013/01/is-there-really-a-spectrum-cri-2007657.pdf
[4] https://techpolicyinstitute.org/press_release/tpi-aspen-forum-working-group-refarming-spectrum-what-have-we-learned/
[5] https://www.congress.gov/112/plaws/publ96/PLAW-112publ96.pdf
[6] http://wireless.fcc.gov/auctions/conferences/combin2003/papers/masterevanjohn.pdf
[7] https://apps.fcc.gov/edocs_public/attachmatch/FCC-15-99A1.pdf, https://transition.fcc.gov/Bureaus/Engineering_Technology/Documents/bulletins/oet63/oet63rev.pdf
[8] https://www.qualcomm.com/news/onq/2015/11/04/how-different-are-lte-u-and-laa
[9] https://blogs.loc.gov/law/2016/02/anniversary-of-the-radio-act-of-1927-the-beginning-of-broadcast-regulation/
[10] https://yalebooks.yale.edu/book/9780300210507/political-spectrum
[11] https://www.ft.com/content/bfffd9fa-e9e2-11de-ae43-00144feab49a
[12] https://auctiondata.fcc.gov/public/projects/1000
[13] https://www.bna.com/unencumbered-spectrum-licenses-n12884909321/
[14] https://www.gao.gov/products/GAO-18-71
[15] https://www.hudson.org/research/13502-the-economic-value-of-property-rights-concepts-in-spectrum-both-with-and-without-licenses
[16] http://iep.clemson.edu/wp-content/uploads/2006/09/Hazlett_IEEE_2006.pdf
[17] https://scholarship.law.berkeley.edu/cgi/viewcontent.cgi?article=1976&context=btlj
[18] https://scholarship.law.berkeley.edu/cgi/viewcontent.cgi?article=1900&context=btlj
